Manual
MOE2014
MOE with MMFF94X forcefield
MOE for protonation state calculation at pH 4.5
MOE was used to add hydrogens and forcefield parameters.
None
After searched the PDBbind-CN database (http://www.pdbbind-cn.org/), we found that all ligands in the absolute free energy calculation subchallenge are similar to the ligand in PDB ID 3K5C.
So we just manually modified the ligand in 3K5C and minimized the added atom groups with MOE to build the initial complex structures.
Yes
Yes
AS-IE
Amber16, AmberTools17, AS-IE (developed by our own laboratory, DOI:10.1021/acs.jctc.7b01295)
ff14SB forcefield for protein, GAFF forcefield with AM1-bcc charges for ligand
energy minimization in 30000 steps, heat to 300.0K in 300ps, density equilibration in 1ns, 20ns production
5.0 kcal/mol/A^2 restraints on protein backbone
last 10ns production trajectories for AS-IE with 50000 snapshots for ASIE and 200 snapshots for ASGB
OBC GB model I for polar solvation energy, 0.0050 kcal/mol/A^2 surften with LCPO SASA for nonpolar solvation energy
residues near 5.0A from ligand, 1.0,3.0,10.0 internal dielectric for nonpolar,polar,charged residues
the VDW epsilon parameters were scaled by 0.09 for ring sp2 carbon and methyl sp3 carbon on protein residues
First, molecular dynamics simulations (MD) were performed on each complex.
ff14SB forcefield was used for protein, and GAFF forcefield with AM1-bcc charges was used for ligands.
TIP3P water boxes were used, and counterions Na+ or Cl- were added to make the system electrically neutral.
Energy minimizations with 30000 steps were performed followed by heating to 300.0K in 300ps and density equilibration under 1.0bar in 1ns.
20ns productions were then performed under 300.0K and 1.0bar, 50000 complex snapshots were recorded during the last 10ns and used for free energy calculations.
During MD, 5.0 kcal/mol/A^2 restraints were added on protein backbone, nonbond cutoff was set to 10.0A, the SHAKE method was used for bonds involving H atoms, and time step was set to 2fs.
Then, each trajectory was analyzed by the AS-IE method.
In this method, the total binding free energy was approximated by adding the binding contribution of each pocket residue within 5A from the ligand.
The binding contribution of one pocket residue to ligand was divided into the enthalpy contribution and the entropy contribution.
The enthalpy contribution of one pocket residue was estimated by the difference between the interaction enthalpy calculated by MMGBSA (GB) of receptor/ligand and receptor'/ligand, where in receptor' the pocket residue was mutated to alanine.
The entropy contribution of one pocket residue was estimated by the difference between the interaction entropy (IE) of the pocket-residue/ligand and alanine/ligand.
50000 snapshots were used by ASIE to calculate the entropy change, and 200 snapshots were used by ASGB to get the enthalpy change.
The OBC GB model I was used to calculate the polar solvation energy, and LCPO SASA method with surften set to 0.0050 kcal/mol/A^2 was used for nonpolar solvation energy during GB.
The system internal dielectric was set based on the type of the residue that alanined, with 1.0 for nonpolar residue, 3.0 for polar residue, and 10.0 for charged residue.
The VDW epsilon parameters were scaled by 0.09 for ring sp2 carbon and methyl sp3 carbon on protein residues.
For each complex, the result was averaged from three independent MD trajectories.
No
No